serial: imx: support RS-485 Rx disable on Tx
diff mbox

Message ID 3b5ba06fdb9c1bdd0b3018bf2f623f52b2856d18.1456651551.git.baruch@tkos.co.il
State New, archived
Headers show

Commit Message

Baruch Siach Feb. 28, 2016, 9:25 a.m. UTC
Some RS-232 to RS-485 transceivers require Rx to be disabled on Tx to
avoid echo of Tx data into the Rx buffer. Specifically, the XR3160E
RS-232/RS-485/RS-422 transceiver behaves this way.

This commit disables Rx on active Tx when SER_RS485_ENABLED is active and
SER_RS485_RX_DURING_TX is disabled.

Note that this is a change in behavior of the driver. Until now
SER_RS485_RX_DURING_TX was enabled unconditionally even when disabled in
the TIOCSRS485 ioctl serial_rs485 flags field.

Cc: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
---
 drivers/tty/serial/imx.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

Comments

Uwe Kleine-König Feb. 28, 2016, 9:56 a.m. UTC | #1
Hello Baruch,

On Sun, Feb 28, 2016 at 11:25:51AM +0200, Baruch Siach wrote:
> Some RS-232 to RS-485 transceivers require Rx to be disabled on Tx to
> avoid echo of Tx data into the Rx buffer. Specifically, the XR3160E
> RS-232/RS-485/RS-422 transceiver behaves this way.
> 
> This commit disables Rx on active Tx when SER_RS485_ENABLED is active and
> SER_RS485_RX_DURING_TX is disabled.
> 
> Note that this is a change in behavior of the driver. Until now

But this change is a good one (assuming it does what it advertises :-).
Userspace got informed before that SER_RS485_RX_DURING_TX is enabled, so
this is not an incompatible change.

Best regards
Uwe

> SER_RS485_RX_DURING_TX was enabled unconditionally even when disabled in
> the TIOCSRS485 ioctl serial_rs485 flags field.
> 
> Cc: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
> Signed-off-by: Baruch Siach <baruch@tkos.co.il>
> ---
>  drivers/tty/serial/imx.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/tty/serial/imx.c b/drivers/tty/serial/imx.c
> index 9362f54c816c..333d34ff358c 100644
> --- a/drivers/tty/serial/imx.c
> +++ b/drivers/tty/serial/imx.c
> @@ -361,6 +361,8 @@ static void imx_stop_tx(struct uart_port *port)
>  			imx_port_rts_inactive(sport, &temp);
>  		else
>  			imx_port_rts_active(sport, &temp);
> +		if (!(port->rs485.flags & SER_RS485_RX_DURING_TX))
> +			temp |= UCR2_RXEN;
>  		writel(temp, port->membase + UCR2);
>  
>  		temp = readl(port->membase + UCR4);
> @@ -568,6 +570,8 @@ static void imx_start_tx(struct uart_port *port)
>  			imx_port_rts_inactive(sport, &temp);
>  		else
>  			imx_port_rts_active(sport, &temp);
> +		if (!(port->rs485.flags & SER_RS485_RX_DURING_TX))
> +			temp &= ~UCR2_RXEN;
>  		writel(temp, port->membase + UCR2);

Can this happen:

 - SER_RS485_RX_DURING_TX is off
 - thread A starts sending (and so disables RX)
 - thread B sets SER_RS485_RX_DURING_TX
 - thread A finishes sending, and doesn't restore RXEN.

?

Even if this cannot happen it might be more robust to restore RXEN
unconditionally in imx_stop_tx?!

Best regards
Uwe
Baruch Siach Feb. 28, 2016, 10:23 a.m. UTC | #2
Hi Uwe,

Thanks for your prompt response.

On Sun, Feb 28, 2016 at 10:56:01AM +0100, Uwe Kleine-König wrote:
> On Sun, Feb 28, 2016 at 11:25:51AM +0200, Baruch Siach wrote:
> > Some RS-232 to RS-485 transceivers require Rx to be disabled on Tx to
> > avoid echo of Tx data into the Rx buffer. Specifically, the XR3160E
> > RS-232/RS-485/RS-422 transceiver behaves this way.
> > 
> > This commit disables Rx on active Tx when SER_RS485_ENABLED is active and
> > SER_RS485_RX_DURING_TX is disabled.
> > 
> > Note that this is a change in behavior of the driver. Until now
> 
> But this change is a good one (assuming it does what it advertises :-).
> Userspace got informed before that SER_RS485_RX_DURING_TX is enabled, so
> this is not an incompatible change.

I thought it is a good idea to mention this fact in the commit log anyway. It 
is not hard to imagine broken userspace being affected by this change.

> > SER_RS485_RX_DURING_TX was enabled unconditionally even when disabled in
> > the TIOCSRS485 ioctl serial_rs485 flags field.
> > 
> > Cc: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
> > Signed-off-by: Baruch Siach <baruch@tkos.co.il>
> > ---
> >  drivers/tty/serial/imx.c | 5 ++++-
> >  1 file changed, 4 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/tty/serial/imx.c b/drivers/tty/serial/imx.c
> > index 9362f54c816c..333d34ff358c 100644
> > --- a/drivers/tty/serial/imx.c
> > +++ b/drivers/tty/serial/imx.c
> > @@ -361,6 +361,8 @@ static void imx_stop_tx(struct uart_port *port)
> >  			imx_port_rts_inactive(sport, &temp);
> >  		else
> >  			imx_port_rts_active(sport, &temp);
> > +		if (!(port->rs485.flags & SER_RS485_RX_DURING_TX))
> > +			temp |= UCR2_RXEN;
> >  		writel(temp, port->membase + UCR2);
> >  
> >  		temp = readl(port->membase + UCR4);
> > @@ -568,6 +570,8 @@ static void imx_start_tx(struct uart_port *port)
> >  			imx_port_rts_inactive(sport, &temp);
> >  		else
> >  			imx_port_rts_active(sport, &temp);
> > +		if (!(port->rs485.flags & SER_RS485_RX_DURING_TX))
> > +			temp &= ~UCR2_RXEN;
> >  		writel(temp, port->membase + UCR2);
> 
> Can this happen:
> 
>  - SER_RS485_RX_DURING_TX is off
>  - thread A starts sending (and so disables RX)
>  - thread B sets SER_RS485_RX_DURING_TX
>  - thread A finishes sending, and doesn't restore RXEN.
> 
> ?
> 
> Even if this cannot happen it might be more robust to restore RXEN
> unconditionally in imx_stop_tx?!

Sounds like a good idea. But if I take your comment to its logical conclusion, 
thread B might just disable SER_RS485_ENABLED entirely. Would it make sense to 
restore RXEN outside the 'if (port->rs485.flags & SER_RS485_ENABLED)' block?  
Or maybe we should just set RXEN in imx_rs485_config() when 
SER_RS485_RX_DURING_TX is enabled?

baruch
Uwe Kleine-König Feb. 29, 2016, 8:51 a.m. UTC | #3
Hello Baruch,

On Sun, Feb 28, 2016 at 12:23:23PM +0200, Baruch Siach wrote:
> Thanks for your prompt response.
> 
> On Sun, Feb 28, 2016 at 10:56:01AM +0100, Uwe Kleine-König wrote:
> > On Sun, Feb 28, 2016 at 11:25:51AM +0200, Baruch Siach wrote:
> > > Some RS-232 to RS-485 transceivers require Rx to be disabled on Tx to
> > > avoid echo of Tx data into the Rx buffer. Specifically, the XR3160E
> > > RS-232/RS-485/RS-422 transceiver behaves this way.
> > > 
> > > This commit disables Rx on active Tx when SER_RS485_ENABLED is active and
> > > SER_RS485_RX_DURING_TX is disabled.
> > > 
> > > Note that this is a change in behavior of the driver. Until now
> > 
> > But this change is a good one (assuming it does what it advertises :-).
> > Userspace got informed before that SER_RS485_RX_DURING_TX is enabled, so
> > this is not an incompatible change.
> 
> I thought it is a good idea to mention this fact in the commit log anyway. It 
> is not hard to imagine broken userspace being affected by this change.
> 
> > > SER_RS485_RX_DURING_TX was enabled unconditionally even when disabled in
> > > the TIOCSRS485 ioctl serial_rs485 flags field.
> > > 
> > > Cc: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
> > > Signed-off-by: Baruch Siach <baruch@tkos.co.il>
> > > ---
> > >  drivers/tty/serial/imx.c | 5 ++++-
> > >  1 file changed, 4 insertions(+), 1 deletion(-)
> > > 
> > > diff --git a/drivers/tty/serial/imx.c b/drivers/tty/serial/imx.c
> > > index 9362f54c816c..333d34ff358c 100644
> > > --- a/drivers/tty/serial/imx.c
> > > +++ b/drivers/tty/serial/imx.c
> > > @@ -361,6 +361,8 @@ static void imx_stop_tx(struct uart_port *port)
> > >  			imx_port_rts_inactive(sport, &temp);
> > >  		else
> > >  			imx_port_rts_active(sport, &temp);
> > > +		if (!(port->rs485.flags & SER_RS485_RX_DURING_TX))
> > > +			temp |= UCR2_RXEN;
> > >  		writel(temp, port->membase + UCR2);
> > >  
> > >  		temp = readl(port->membase + UCR4);
> > > @@ -568,6 +570,8 @@ static void imx_start_tx(struct uart_port *port)
> > >  			imx_port_rts_inactive(sport, &temp);
> > >  		else
> > >  			imx_port_rts_active(sport, &temp);
> > > +		if (!(port->rs485.flags & SER_RS485_RX_DURING_TX))
> > > +			temp &= ~UCR2_RXEN;
> > >  		writel(temp, port->membase + UCR2);
> > 
> > Can this happen:
> > 
> >  - SER_RS485_RX_DURING_TX is off
> >  - thread A starts sending (and so disables RX)
> >  - thread B sets SER_RS485_RX_DURING_TX
> >  - thread A finishes sending, and doesn't restore RXEN.
> > 
> > ?
> > 
> > Even if this cannot happen it might be more robust to restore RXEN
> > unconditionally in imx_stop_tx?!
> 
> Sounds like a good idea. But if I take your comment to its logical conclusion, 
> thread B might just disable SER_RS485_ENABLED entirely. Would it make sense to 
> restore RXEN outside the 'if (port->rs485.flags & SER_RS485_ENABLED)' block?  
> Or maybe we should just set RXEN in imx_rs485_config() when 
> SER_RS485_RX_DURING_TX is enabled?

The latter sounds like the right thing to do.

Best regards
Uwe

Patch
diff mbox

diff --git a/drivers/tty/serial/imx.c b/drivers/tty/serial/imx.c
index 9362f54c816c..333d34ff358c 100644
--- a/drivers/tty/serial/imx.c
+++ b/drivers/tty/serial/imx.c
@@ -361,6 +361,8 @@  static void imx_stop_tx(struct uart_port *port)
 			imx_port_rts_inactive(sport, &temp);
 		else
 			imx_port_rts_active(sport, &temp);
+		if (!(port->rs485.flags & SER_RS485_RX_DURING_TX))
+			temp |= UCR2_RXEN;
 		writel(temp, port->membase + UCR2);
 
 		temp = readl(port->membase + UCR4);
@@ -568,6 +570,8 @@  static void imx_start_tx(struct uart_port *port)
 			imx_port_rts_inactive(sport, &temp);
 		else
 			imx_port_rts_active(sport, &temp);
+		if (!(port->rs485.flags & SER_RS485_RX_DURING_TX))
+			temp &= ~UCR2_RXEN;
 		writel(temp, port->membase + UCR2);
 
 		/* enable transmitter and shifter empty irq */
@@ -1614,7 +1618,6 @@  static int imx_rs485_config(struct uart_port *port,
 	/* unimplemented */
 	rs485conf->delay_rts_before_send = 0;
 	rs485conf->delay_rts_after_send = 0;
-	rs485conf->flags |= SER_RS485_RX_DURING_TX;
 
 	/* RTS is required to control the transmitter */
 	if (!sport->have_rtscts)